Document Type

Conference Paper

Publication Date

2026

Abstract

The rapid adoption of generative AI tools has fundamentally challenged traditional assessments in higher education, particularly in large-enrollment and fully asynchronous online courses where instructor presence and grading time are constrained. While many instructors have responded by restricting AI use or adopting detection tools, research increasingly shows that detection-based approaches are unreliable and labor-intensive. We propose a design-centered alternative: reshaping assessments so that AI use becomes insufficient, rather than prohibited.

Drawing on a systematic process of testing proposed and original assessment ideas directly against multiple generative AI models (using both simple and advanced prompting strategies), we present a curated playbook of AI-resistant activities that function within a standard LMS and scale effectively to large or online courses. These activities can preserve necessary foundational learning while requiring students to exercise disciplinary judgment, make context-bound decisions, and engage in application that cannot be satisfied through single-prompt or copy-and-paste AI outputs.

Download

Share

COinS